Up to 200 scientists picked through the European Commission's RAISE program will get access to frontier AI models, including AlphaFold 3 and WeatherNext
Google DeepMind and Google Cloud are opening their AI toolbox to European science. Up to 200 researchers across the continent will get access to frontier AI models and a shared pool of $3 million in compute credits.
The announcement came on October 14, 2026, at an AI summit in Dublin. The European Commission, not Google, picks who gets in.
What researchers actually get
The offer has two parts: access to Google’s most advanced AI models, and the cloud computing power needed to run them.
The model lineup includes AlphaFold 3, known for precise protein structure prediction, and WeatherNext, built for advanced weather forecasting. Rounding out the list are AlphaGenome, AlphaEvolve, and AlphaEarth Foundations. Together, the tools are pitched at fields like molecular mapping, weather forecasting, and genomic analysis.
If the $3 million were split evenly across all 200 slots, each researcher would get about $15,000 in credits.
The RAISE connection
Participants will be drawn from the Resource for AI Science in Europe, known as RAISE. The European Commission oversees the program. Its goal is to boost AI-driven scientific work across Europe.
This setup gives the initiative a public-sector filter. Google provides the models and the compute. Brussels decides who uses them.
The stated aim is to tackle a familiar headache. Researchers often struggle to get the advanced computing resources that modern AI work demands.
What this means
For European science, the most direct effect is practical. Up to 200 research teams get tools and computing power they may not have been able to afford. Fields like protein research, climate and weather modeling, and genomics stand to benefit most, given the models on offer.
$3 million is meaningful for individual labs. It is modest relative to the size of European research overall.
For the European Commission, the partnership offers speed. Building sovereign AI infrastructure takes years. Plugging researchers into existing tools takes months. The tradeoff is reliance on a US technology company for core scientific capabilities.
